Initial web research, with the public record and case status checked September 27, 2026.
Case record and current status
The CFPB issued separate Apple and Goldman Sachs Bank USA orders on October 23, 2024. Apple’s case concerned dispute transmission and installment-enrollment practices. Apple consented without admitting or denying the findings, except jurisdiction. The Bureau’s findings are agency findings in a consent order, not a trial verdict.
On September 22, 2025, the CFPB terminated Apple’s order and waived alleged noncompliance with that order. The termination document states that Apple paid its $25 million civil penalty. This status was checked September 27, 2026. The Apple termination does not terminate Goldman’s separate order; the Goldman case page reviewed lists its original order and does not establish a comparable termination.
How the handoff failed in the agency’s account
Apple provided the customer-facing interface and Goldman extended credit and investigated disputes. According to the Apple order, a second form introduced in 2020 resulted in tens of thousands of disputes not being passed along when consumers did not finish that extra step. The agency also found that marketing and checkout design could lead consumers to expect interest-free monthly installments without being properly enrolled.
The Goldman case record describes separate findings involving investigation, notices, credit reporting and installment representations. Its order required $19.8 million of consumer redress and a $45 million civil penalty. These amounts and duties belong to Goldman’s action; they should not be attributed to Apple or combined into an assertion that Apple’s terminated order remains active.
Operating analysis: customer intent must survive system boundaries
A product can record a customer’s request in one interface while the legally responsible workflow never receives it. An aggregate application-availability metric will not expose that failure. A stronger control follows an individual request across intake, transmission, acceptance, investigation and resolution. Each system should retain a common identifier and a reconcilable timestamp.
This suggests an operational test for bank partnerships: reconcile the number of customer requests at the interface with accepted records at the servicing system, investigate the difference, and confirm how abandoned or partially completed flows are handled. Merely testing the ideal completed journey misses the customers most likely to need help. These are suggested control designs, not a restatement of an active Apple order.
Illustrative test cases for a card or installment launch
A hypothetical launch review should include a consumer who submits the first dispute screen and exits, a network interruption after submission, a duplicate request, an account with both promotional and regular balances, and a refund after an installment purchase. The objective is to prove the resulting treatment, not only that every button works.
| Journey | Question to resolve before launch |
|---|---|
| Dispute intake | Does the responsible investigator receive and acknowledge the request? |
| Incomplete follow-up | Is the original request preserved and routed appropriately? |
| Installment selection | Can the customer tell whether the purchase actually entered the plan? |
| Mixed balances and refund | Does the disclosed treatment match the account’s actual calculation? |
Commercial implications, competing view and monitoring
For merchants and lenders, frictionless checkout has value, but ambiguous enrollment can create servicing cost, complaint volume and customer harm downstream. Measure completed, correctly understood enrollment rather than conversion alone. For disputes, a lower contact count can reflect customers giving up instead of better outcomes. Pair efficiency measures with reconciliation and outcome testing.
It would also be wrong to assume an old consent order proves today’s interface still has the same defect. Apple’s termination is a material part of the record. Current performance would require current evidence. Future agency dispositions, verified workflow changes and supported complaint or redress data could change the view. The reviewed public sources do not establish the present performance of every Apple Card process or resolve all other parties’ obligations.